Point-to-Point Learning and Tracking for Networked Stochastic Systems With Fading Communications
针对通信信号衰落导致信息偏差的网络化随机系统，提出一种带递减增益序列的迭代学习控制方案，仅需部分位置高精度跟踪，并用信道统计直接校正，严格证明了收敛性。
This article studies the point-to-point (P2P) learning and tracking problem for networked stochastic systems with fading communications by iterative learning control. The P2P tracking problem indicates that only partial positions rather than the whole reference are required to achieve high tracking precision. An auxiliary matrix is introduced to connect the entire reference and the required tracking targets. The fading communication introduces multiplicative randomness to the transmitted signals, which leads to the biased available information. A direct correction mechanism is employed using statistics of the communication channel. A learning control scheme is then proposed with a decreasing gain sequence to ensure steady convergence in the presence of various types of randomness. Two scenarios of varying initial states are considered. The convergence of the proposed scheme is strictly established. The validity of the proposed algorithm is verified by two simulation examples.